Quantifying health effects of cannabis use and cannabis use disorder on geriatric syndromes and HIV outcomes: causal effects analysis in the Veterans Aging Cohort Study and the Mt. Sinai HIV Cohort

About This Grant

PROJECT SUMMARY/ABSTRACT Shifting cannabis policy and access to high potency cannabis products creates an urgent need to evaluate and disseminate data on cannabis exposure and risks among aging people living with HIV (PWH). Higher potency products confer heightened risk of cannabis use disorder (CUD), cognitive impairment, and falls. Because PWH are more frail than people without HIV (PWoH), they may be more susceptible cannabis-associated harms (e.g., geriatric syndromes including falls/fractures, delirium/dementia) and ART failure or non-adherence. Our proposal responds to RFA-DA-26-055 (Accelerating the Pace of Drug Abuse Research Using Existing Data) and NIDA priorities of understanding the role of substance use in HIV pathogenesis. Electronic health record (EHR) data can advance cannabis research; innovative approaches are needed to maximize its utility. We will use longitudinal nested Veterans Aging Cohort Study (VACS) cohorts and our expertise in informatics, cannabis epidemiology/policy research, HIV, and causal inference to estimate causal effects of cannabis use on geriatric syndromes and ART failure/adherence. Specific Aims: Aim 1. Determine the association of CUD and incident geriatric syndromes and ART failure/adherence. In VACS-HIV (N~180,000), we will estimate the association of CUD and incident falls and fractures (Aim 1Ai) and delirium and dementia (Aim 1Aii). Among PWH, we will determine the association of CUD and HIV viral suppression and ART adherence (Aim 1B). We will assess how associations (Aim 1A-1B) generalize to a non-veteran population (Mt. Sinai HIV cohort, N~850,000, Aim 1C-D). We will explore how cannabis legalization alters these associations (Aim 1E). Aim 2. Estimate causal effect of cannabis use (self-reported) on incident geriatric syndromes. In VACS-Survey (N~9000), we will use target trial emulation to determine the effect of past-year cannabis use on incident falls and fractures (Aim 2Ai) and delirium and dementia (Aim 2Aii). Aim 3. Build and validate a novel EHR-based cannabis use ascertainment algorithm (NLPWH) by extending an existing natural language processing algorithm (NLPREOP) and validating NLPWH. We will create NLPWH by retraining NLPREOP on multiple EHR clinical note types (e.g., primary care) from VACS-HIV in 9 U.S. states (Aim 3A); internally validate NLPWH using cannabis biomarker and self-report (Aim 3B); externally validate NLPWH by chart review in the Mt. Sinai HIV cohort (Aim 3C); and assess impact of cannabis legalization on algorithm performance (Aim 3D). Using the target trial approach from Aim 2, we will evaluate the causal association of NLPWH-derived cannabis use with geriatric syndromes, ART failure, and ART adherence. IMPACT: Using data from two healthcare systems we will provide early and increasingly complete evidence identifying and quantifying medical harms from cannabis use among PWH and PWoH. The new NLP tool we validate may help health systems rapidly estimate cannabis exposure in their patients and guide urgently needed policies for CUD screening, diagnosis, and treatment.